The present invention relates to data transmission in communication systems, such as the recording and reproduction of binary data in disk storage systems for digital computers, particularly to a sampled amplitude read channel employing a 2, 2, 1 asymmetric partial response target which requires less equalization to attain the desired dipulse response.
Disk storage systems (magnetic and optical disk drives) are essentially communication systems wherein the media, heads, and read/write electronics constitute the communication channel. Disk storage systems have transitioned from simple analog pulse detector read channels to more sophisticated sampled amplitude read channels comprising complex sequence detectors operating on synchronous, discrete-time samples of the read signal. Sequence detectors generate an estimated data sequence by evaluating several read signal sample values in context to estimate the most likely recorded data sequence to have generated the read signal sample values. A Viterbi sequence detector, for example, selects an estimated data sequence which minimizes a Euclidean distance between the read signal samples and ideal sample values corresponding to the possible recorded data sequences. The accuracy of a sequence detector in selecting the correct estimated data sequence depends on the magnitude and character of the noise in the read signal.
A Viterbi sequence detector is a maximum likelihood sequence detector (optimal) only if the noise is statistically independent with a Gaussian probability distribution. This is typically not the case in disk storage systems due to the equalizers in the read channel which correlate the noise in the read signal. The read signal is typically equalized to conform to a desired partial response (PR) which means that an isolated symbol will generate a particular dipulse shape which minimizes intersymbol interference. The recording channel""s natural dipulse response (unequalized response) is typically very wide and possibly asymmetric. If the read signal were not equalized, there would be several samples in the dipulse response resulting in an unwieldy large state transition diagram for the sequence detector. For this reason the read signal is equalized to reduce the number of samples in the dipulse response, but the noise correlating effect of the channel equalizer""s degrades the performance of the sequence detector.
There are various techniques employed in the prior art which attempt to compensate for the undesirable noise correlating effect of the channel equalizers. For example, it is known to select a wider PR target (e.g., EPR4 over PR4) so that less equalization is required to match the channel""s response to the target response. As described above, however, a wider dipulse response increases the cost and complexity of the resulting state transition diagram for the sequence detector. And although the noise correlating effect is reduce, it is not eliminated.
Other techniques have also been employed, such as noise predictive algorithms which augment the sequence detector by incorporating the noise correlating effect of the channel equalizers into the sequence detector""s state transition diagram, but this increases the cost and complexity of the sequence detector. Noise whitening filters may also be employed in post processors which attempt to whiten the noise in the read signal before detecting and correcting the dominant error events associated with the sequence detector. Although these techniques improve performance, further gains are attainable.
There is, therefore, a need to compensate for the noise correlating effect of channel equalizers in a sampled amplitude read channel for disk storage systems. In particular, there is a need to minimize the noise correlating effect of the channel equalizers in order to enhance the sequence detector""s performance without significantly increasing the cost and complexity of the read channel.
The present invention may be regarded as a sampled amplitude read channel for reading data recorded on a disk storage medium by detecting an estimated data sequence from a sequence of read signal sample values generated by sampling an analog read signal emanating from a read head positioned over the disk storage medium. A sampling device samples the analog read signal to generate the read signal sample values, and a discrete-time equalizer equalizes the read signal sample values according to an asymmetric partial response target comprising a dipulse response of the form:
( . . . , 0, 0,+X0,+X1,xe2x88x92X2,xe2x88x92X3,xe2x88x92X4, 0, 0, . . . )
where X0xe2x88x92X4 are non-zero to thereby generate equalized sample values. In the embodiments disclosed herein, X0xe2x88x92X4 are 2, 2, 1, 2, 1 respectively. A discrete-time sequence detector detects the estimated data sequence from the equalized sample values.
The present invention may also be regarded as a method of reading data recorded on a disk storage medium by detecting an estimated data sequence from a sequence of read signal sample values generated by sampling an analog read signal emanating from a read head positioned over the disk storage medium. The analog read signal is sampled to generate the read signal sample values which are equalized according to an asymmetric partial response target comprising a dipulse response of the form:
( . . . , 0, 0,+X0,+X1,xe2x88x92X2,xe2x88x92X3,xe2x88x92X4, 0, 0, . . . )
where X0xe2x88x92X4 are non-zero to thereby generate equalized sample values.
In the embodiments disclosed herein, X0xe2x88x92X4 are 2, 2, 1, 2, 1 respectively. An estimated data sequence is detected from the equalized sample values.